check_point_location() R function from [disprofas]

Check point location

The function check_point_location() checks if points that were found by the gep_by_nera() function sit on specified confidence region bounds ( $\textit{CRB}$ ) or not. This is necessary because the points found by aid of the Method of Lagrange Multipliers (MLM) and Newton-Raphson (nera) optimisation may not sit on the $\textit{CRB}$ .


check_point_location(lpt, lhs)

Arguments

lpt: A list returned by the gep_by_nera() function.
lhs: A list of the estimates of Hotelling's two-sample $T^2$

statistic for small samples as returned by the function get_T2_two().

Returns

The function returns the list that was passed in via the lpt

parameter with a modified points.on.crb element, i.e. set as TRUE if the points sit on the $\textit{CRB}$ or FALSE if they do not sit on the $\textit{CRB}$ .

Details

The function check_point_location() checks if points that were found by the gep_by_nera() function sit on specified confidence region bounds ( $\textit{CRB}$ ) or not. The gep_by_nera() function determines the points on the $\textit{CRB}$ for each of the $n_p$ time points or model parameters by aid of the Method of Lagrange Multipliers (MLM) and by Newton-Raphson (nera) optimisation, as proposed by Margaret Connolly (Connolly 2000). However, since the points found may not sit on the specified $\textit{CRB}$ , it must be checked if the points returned by the gep_by_nera() function do sit on the $\textit{CRB}$

or not.

Examples


# Collecting the required information
time_points <- suppressWarnings(as.numeric(gsub("([^0-9])", "",
                                                colnames(dip1))))
tcol <- which(!is.na(time_points))
b1 <- dip1$type == "R"
tol <- 1e-9

# Hotelling's T2 statistics
l_hs <- get_T2_two(m1 = as.matrix(dip1[b1, tcol]),
                   m2 = as.matrix(dip1[!b1, tcol]),
                   signif = 0.05)

# Calling gep_by_nera()
res <- gep_by_nera(n_p = as.numeric(l_hs[["Parameters"]]["df1"]),
                   kk = as.numeric(l_hs[["Parameters"]]["K"]),
                   mean_diff = l_hs[["means"]][["mean.diff"]],
                   m_vc = l_hs[["S.pool"]],
                   ff_crit = as.numeric(l_hs[["Parameters"]]["F.crit"]),
                   y = rep(1, times = l_hs[["Parameters"]]["df1"] + 1),
                   max_trial = 100, tol = tol)

# Expected result in res[["points.on.crb"]]
# [1] NA

# Check if points lie on the confidence region bounds (CRB)
check_point_location(lpt = res, lhs = l_hs)

# Expected result in res[["points.on.crb"]]
# [1] TRUE

References

Tsong, Y., Hammerstrom, T., Sathe, P.M., and Shah, V.P. Statistical assessment of mean differences between two dissolution data sets. Drug Inf J. 1996; 30 : 1105-1112.

tools:::Rd_expr_doi("10.1177/009286159603000427")

Connolly, M. SAS(R) IML Code to calculate an upper confidence limit for multivariate statistical distance; 2000; Wyeth Lederle Vaccines, Pearl River, NY.

https://analytics.ncsu.edu/sesug/2000/p-902.pdf

check_point_location function